bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Defining the RBPome of T helper cells to study higher order post-transcriptional gene
regulation
Kai P. Hoefig1*, Alexander Reim2*, Christian Gallus3*, Elaine H. Wong4, Gesine Behrens1,
Christine Conrad4, Meng Xu1, Taku Ito-Kureha4, Kyra Defourny4,10, Arie Geerlof5, Josef
Mautner6, Stefanie M. Hauck7, Dirk Baumjohann4,11, Regina Feederle8, Matthias Mann2,
Michael Wierer2,12#, Elke Glasmacher3,9#, Vigo Heissmeyer1,4#
1 Research Unit Molecular Immune Regulation, Helmholtz Center Munich, Munich, Germany
2 Department of Proteomics and Signal Transduction, Max-Planck-Institute of Biochemistry,
Munich, Germany
3 Institute of Diabetes and Obesity, Helmholtz Center Munich, Munich, Germany
4 Institute for Immunology at the Biomedical Center, Ludwig-Maximilians-Universität
München, Planegg-Martinsried, Germany
5 Institute of Structural Biology, Helmholtz Center Munich, Neuherberg, Germany
6 Research Unit Gene Vectors, Helmholtz Center Munich & Children’s Hospital, TU Munich,
Germany
7 Research Unit Protein Science, Helmholtz Center Munich, Munich, Germany
8 Monoclonal Antibody Core Facility and Research Group, Institute for Diabetes and
Obesity, Helmholtz Center Munich, Neuherberg, Germany.
9 present address: Roche Pharma Research and Early Development, Large Molecule
Research, Roche Innovation Center Munich, Penzberg, Germany
10 present address: Department of Biomolecular Health Sciences, Utrecht University,
Utrecht, the Netherlands
11 present address: Medical Clinic III for Oncology, Immuno-Oncology and Rheumatology
University Hospital Bonn, University of Bonn, Bonn, Germany
1 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
12 present address: Proteomics Research Infrastructure, University of Copenhagen,
Copenhagen, Denmark
* These authors contributed equally
# Corresponding Authors: [email protected]
One sentence statement:
We provide an atlas of RNA-binding proteins in human and mouse T helper cells as a
resource for studying higher order post-transcriptional gene regulation.
Abstract
Post-transcriptional gene regulation is complex, dynamic and ensures proper T cell function.
The targeted transcripts can simultaneously respond to various factors as evident for Icos,
an mRNA regulated by several RNA binding proteins (RBPs), including Roquin. However,
fundamental information about the entire RBPome involved in post-transcriptional gene
regulation in T cells is lacking. Here, we applied global RNA interactome capture (RNA-IC)
and orthogonal organic phase separation (OOPS) to human and mouse primary T cells and
identified the core T cell RBPome. This defined 798 mouse and 801 human proteins as
RBPs, unexpectedly containing signaling proteins like Stat1, Stat4 and Vav1. Based on the
vicinity to Roquin-1 in proximity labeling experiments, we selected ~50 RBPs for testing
coregulation of Roquin targets. Induced expression of these candidate RBPs in wildtype and
Roquin-deficient T cells unraveled several Roquin-independent contributions, but also
revealed Celf1 as a new Roquin-1-dependent and target-specific coregulator of Icos.
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
T lymphocytes as central entities of the adaptive immune system must be able to make
critical cell fate decisions fast. To exit quiescence, commit to proliferation and differentiation,
exert effector functions or form memory they strongly depend on programs of gene
regulation. Accordingly, they employ extensive post-transcriptional regulation through RBPs
and miRNAs or 3’ end oligo-uridylation and m6A RNA modifications. These RBPs, or RBPs
that recognize the modifications, directly affect the expression of genes by controlling mRNA
stability or translation efficiency 1, 2, 3, 4, 5. The studies in T helper cells have focused on a
small number of RNA-binding proteins, including HuR and TTP/Zfp36l1/Zfp36l2 6, 7, 8, 9 ,
Roquin-1/2 10, 11 and Regnase-1/4 12, 13 as well as some miRNAs like miR-17~92, miR-155,
miR-181, miR-125 or miR-146a 14. Moreover, first evidence for m6A RNA methylation in this
cell type has been provided 4. Underscoring the relevance for the immune system, loss-of-
function of these factors has often been associated with profound alterations in T cell
development or functions which caused immune-related diseases 14, 15, 16, 17. Intriguingly,
many key factors of the immune system have acquired long 3’-UTRs enabling their
regulation by multiple, and often overlapping sets of post-transcriptional regulators 18. Some
RBPs recruit additional co-factors as for example Roquin binding with Nufip2 to RNA 19 and
some of them have antagonistic RBPs like HuR and TTP 20 or Regnase-1 and Arid5a 21.
Such functional or physical interactions together with interdependent binding to the
transcriptome create enormous regulatory potential. The challenge is therefore to integrate
our current knowledge about individual RBPs into concepts of higher order gene regulation
that reflect the interplay of different, and ideally of all cellular RBPs.
A prerequisite for studying higher order post-transcriptional networks is to know cell type
specific RBPomes to account for differential expression and RBP plasticity. To this end
several global methods have been developed over the last decade, revealing a growing
number of RBPs that may even exceed recent estimates of ~7.5% of the human proteome
22. RNA interactome capture (RNA-IC) 23 is one widely used, unbiased technique, however, it
is constricted by design, exclusively identifying proteins binding to polyadenylated RNAs. In
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
contrast, orthogonal organic phase separation (OOPS) analyzes all UV-crosslinked protein-
RNA adducts from interphases after organic phase separation.
The interactions of RBPs with RNA typically involve charge-, sequence- or structure-
dependent interactions and to date over 600 structurally different RNA-binding domains
(RBD) have been identified in canonical RBPs of the human proteome 22. However, global
methods also identified hundreds of non-canonical RBPs, which oftentimes contained
intrinsically disordered regions (IDRs). Surprisingly, as many as 71 human proteins with well-
defined metabolic functions were found to interact with RNA 24 introducing the concept of
“moonlighting”. Depending on availability from their “day job” in metabolism such proteins
also bear the potential to impact on RNA regulation. Recent large-scale approaches have
increased the number of EuRBPDB-listed human RBPs to currently 2949 25 , suggesting that
numerous RNA/RBPs interactions and cell-type specific gene regulations have gone
unnoticed so far.
As a first step towards a global understanding of post-transcriptional gene regulation we
experimentally defined all proteins that can be crosslinked to RNA in T helper cells. RNA-IC
and OOPS identified 310 or 1200 proteins in primary CD4+ T cells interacting with
polyadenylated transcripts or all RNA species, respectively. Importantly, this dataset now
enables the study of higher order gene regulation by determining for example how the
cellular RBPs participate or intervene with post-transcriptional control of targets by individual
RBPs.
Results
Icos exhibits simultaneous and temporal regulation through several RBPs
A prominent example for complex post-transcriptional gene regulation is the mRNA encoding
inducible T-cell costimulator (Icos). It harbors a long 3’-UTR, which responds in a redundant
manner to Roquin-1 and Roquin-2 proteins, is repressed by Regnase-1 and microRNAs, but
also contains TTP binding sites 7, 10, 11, 13, 26, 27, 28. Moreover, the Icos 3’-UTR was proposed to
be modified by m6A methylation 29, which could either attract m6A-specific RBPs with YTH
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
domains 30, recruit or repel other RBPs 31, or interfere with base-pairing and secondary
structure- or miRNA/mRNA-duplex formation 32. Because of transcriptional and post-
transcriptional regulation, Icos expression exhibits a hundred-fold upregulation on the protein
level during T cell activation, which then declines after removal of the TCR stimulus (Fig. 1).
To determine the temporal impact on Icos expression by Roquin, Regnase-1, m6A and
miRNA regulation we analyzed inducible, CD4-specific inactivation of Roquin-1 together with
Roquin-2 (Rc3h1-2), Regnase-1 (Zc3h12a) or Wtap, an essential component of the m6A
methyltransferase complex 33, as well as Dgcr8, which is required for pre-miRNA biogenesis
34. To this end, we performed tamoxifen gavage on mice expressing a Cre-ERT2 knockin
allele in the CD4 locus 35 together with the floxed, Roquin paralogs encoding, Rc3h1 and
Rc3h2 alleles (Fig. 1a-c), Regnase-1 encoding Zc3h12a alleles (Fig. 1d-f) as well as Wtap,
(Fig. 1g-i) and Dgcr8 alleles, (Fig. 1j-l). We isolated CD4+ T cells from these mice and
expanded them for five days. Confirming target deletion on the protein level (Fig. 1c, f, i, l)
we determined a strong negative effect on Icos expression by Roquin and Regnase-1 on
days 2-5 (Fig. 1a-b and d-e), a moderate positive effect of Wtap on days 2-5 (Fig. 1g-h),
and only a small effect of Dgcr8, with an initial tendency of negative (day 1) and later on
positive effects (days 4-5) (Fig. 1j-k). We next asked whether T cell activation affects the
expression levels of known regulators of Icos, as well as other RBPs, to install temporal
compartmentalization. To do so, we monitored the expression of a panel of RBPs in mouse
CD4+ T cells over the same time course (Fig. 1m-q). Indeed, we observed fast upregulation
of RBPs as determined with pan-Roquin, Nufip2 19, Fmrp, Fxr1, Fxr2, pan-TTP/Zfp36 7, pan-
Ythdf (Supplementary Fig. 1a), or Celf1 specific antibodies, but also slower accumulation
as determined with Regnase-1 or Rbms1 specific antibodies (Fig. 1m, o). There was also
downregulation of RBPs as shown with pan-AGO 36 and Cpeb4 specific antibodies (Fig. 1m-
n and p). Of note, we also observed signs of post-translational regulation showing
incomplete or full cleavage of Roquin or Regnase-1 proteins 10, 13, respectively (Fig. 1m),
and the induction of a slower migrating band for Celf1, likely phosphorylation 37 (Fig. 1q).
Factors with the potential to cooperate, as involved for Roquin and Nufip219 or Roquin and 5 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Regnase-110, showed overlapping temporal regulation (Fig. 1m), suggesting reinforcing
effects on the Icos target. Together these data indicate that mRNA targets respond to
simultaneous inputs from several RBPs, which are aligned by dynamic expression and post-
translational regulation to orchestrate redundant, cooperative and antagonistic effects into a
higher order regulation.
The polyadenylated RNA-bound proteome of mouse and human T helper cells
To decipher this post-transcriptional network we set out to determine the RNA-binding
protein signature in primary mouse and human CD4+ T cells. To identify mRNA-binding
+ proteins we first performed mRNA capture experiments on CD4 T cells expanded under TH0
culture conditions (Fig. 2a). Pull-down with oligo-dT beads enabled the enrichment of
mRNA-bound proteins as determined by silver staining, which were increased in response to
preceding UV irradiation of the cells (Supplementary Fig. 1b). Reverse transcription
quantitative PCR (RT-qPCR) confirmed the recovery of specific mRNAs, such as for the
housekeeping genes Hprt and β-actin. Both mRNAs were enriched at least 2-3 fold after UV
crosslink, but there was no detection of non-polyadenylated 18S rRNA (Supplementary Fig.
1c). Focusing on protein recovery, we determined greatly enriched polypyrimidine tract-
binding protein 1 (Ptbp1) RBP compared to the negative control β-tubulin in mRNA capture
experiments using the EL-4 thymoma cell line (Supplementary Fig. 1d). Next we performed
mass spectrometry (MS) on captured proteins from murine and human T cells (Fig. 2b-c).
Quantifying proteins bound to mRNA in crosslinked (CL) versus non-crosslinked (nCL)
samples we defined a total of 312 mouse (Fig. 2b) and 308 human mRNA-binding proteins
(mRBPs) (Fig. 2c) with an overlap of ~70% (Supplementary Table 1), which is in
concordance with the overlap of all listed RBPs for these two species in the eukaryotic RBP
database (http://EuRBPDB.syshospital.org). Gene ontology (GO) analysis identified the term
‘mRNA binding’ as most significantly enriched (Fig. 2d). The top ten GO terms in mouse
were also strongly enriched in the human dataset with comparable numbers of proteins
assigned to the individual GO terms in both species (Fig. 2d). RBPs not only bind RNA by
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
classical RBDs, but RNA-interactions can also map to intrinsically disordered regions (IDRs)
38. Furthermore, low complexity regions (LCRs) have also been reported to be
overrepresented in RBPs 23. Indeed, IDRs (Fig. 2e) and LCRs (Supplementary Fig. 1e)
were strongly enriched protein characteristics of mouse and human RNA-IC-identified
RBPomes. We next wondered how much variation exists in the composition of RBPs
between different T helper cell subsets. Repeating RNA-IC experiments in mouse and
human iTreg cells (Supplementary Fig. 2a), we find an overlap of 96% or 90% with the
respective mouse or human iTreg with Teff RBPomes, suggesting that the same RBPs bind
to the transcriptome in different T helper cell subsets (Table 1). Nevertheless, 47 or 48
proteins were exclusively identified in mouse or human effector T cells, respectively, and 10
or 28 proteins were only found in mouse or human Treg cells, respectively (Supplementary
Fig. 2b). Although these differences could indicate the existence of subset-specific RBPs,
they could also be related to an incomplete assessment of RBPs by the RNA-IC technology.
The global RNA-bound proteome of mouse and human T helper cells
We therefore employed a second RNA-centric method of orthogonal organic phase
separation (OOPS) to extend our definition of the RBPome and cross-validate our results.
Similar to interactome capture, OOPS preserves cellular protein/RNA interactions by UV
crosslinking of intact cells. The physicochemical properties of the resulting aducts direct
them towards the interphase in the organic and aqueous phase partitioning procedure (Fig.
3a). Following several cycles of interphase transfer and phase partitioning, RNase treatment
releases RNA-bound proteins into the organic phase, making them amenable to mass
spectrometry 39, 40. Evaluating the method, we investigated RNA and proteins from purified
interphases derived from CL and nCL MEF cell samples. RNAs with crosslinked proteins
purified from interphases hardly migrated into agarose gels, but regained normal migration
behavior after protease digest, as judged from the typical 18S and 28S rRNA pattern
(Supplementary Fig. 3a). Conversely, known RBPs like Roquin-1 and Gapdh appeared
after crosslinking in the interphase and could be recovered after RNase treatment from the
7 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
organic phase (Supplementary Fig. 3b). Utilizing the same cell numbers and culture
conditions of T cells, this method identified in total 1255 and 1159 significantly enriched
RBPs for mouse or human T cells, respectively, when comparing CL and nCL samples
(Supplementary Table 1). The overlap between both organisms was 55% (Fig. 3b) and
60% (Fig. 3c) in relation to the individual mouse and human RBPomes. Although
glycosylated proteins are known to also accumulate in the interphase 39, we experimentally
verified that they did not migrate into the organic phase after RNase treatment
(Supplementary Fig. 3c-d). Analyzing OOPS-derived RBPomes for gene ontology
enrichment using the same approach as for RNA-IC the GO term ‘mRNA binding’ was again
most significantly enriched in mouse and man (Fig. 3d). The top 10 GO terms were RNA
related and six of them overlapped with those identified for RNA-IC-derived RBPomes. High
similarity between mouse and human RBPomes becomes apparent by the similarity in all
GO categories, including ‘molecular function’ (Fig. 3d), ‘biological process’ and ‘cellular
component’ (Supplementary Fig. 4). Although our OOPS approach exceeded by far the
quantity of RNA-IC identified RBPs, the number of ~1200 RBPs well matched published
RBPomes of HEK293 (1410 RBPs), U2OS (1267 RBPs) and MCF10A (1165 RBPs) cell
lines 39.
Defining the core T helper cell RBPome
To define a T helper cell RBPome we first made sure that RNA-IC and OOPS did not
preferentially identify high abundance proteins (Fig. 4a-b). In comparison to single shot total
proteomes OOPS-identified RBPs spanned the whole range of protein expression without
apparent bias. In general, this was also true for RNA-IC, with a slight tendency to more
abundantly expressed proteins. This however might be a true effect since messenger RNA
binding RBPs have been reported to be higher in expression even in comparison to other
RBPs 22. We used the recently established comprehensive eukaryotic RBP database as
reference to compare OOPS and RNA-IC-identified canonical and non-canonical RBPs from
mouse and human CD4+ T cells. The numbers of proteins in the mouse T helper cell
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
RBPomes created by RNA-IC and OOPS ranging from 312 to 1255 made up 10% to 40% of
all listed EuRBPDB proteins, respectively (Fig. 4c). OOPS-identified T cell RBPs
outnumbered those from RNA-IC experiments by a factor of four, which was predominantly
due to the eight times higher number of non-canonical RBPs. Interestingly though, there
were also twice as many canonical RBPs significantly enriched by OOPS (Fig. 4c).
Analyzing the ten most abundantly annotated mouse RBDs (comprising 26 to 224 members)
showed that RNA-IC and OOPS often identified the same canonical RBPs (Supplementary
Fig. 5), however at least equal, higher or much higher numbers were detected in OOPS
samples depending on the specific RBD (Fig. 4d). These findings suggested that the OOPS
method recovered RBPs from additional, non-polyadenylated RNAs and that RNA-IC-
derived RBPomes are specific but incomplete. These conclusions are also supported by
highly similar results obtained for the human CD4+ T cell RBPome (Fig. 4e-f). In a 4-way
comparison of mouse and human RBPs identified by OOPS and RNA-IC (Fig. 4g) we
conservatively defined all proteins that were identified by at least two datasets as ‘core CD4+
T cell RBPomes’ discovering 798 mouse and 801 human RBPs in this category
(Supplementary Table 1). A sizable number of 519 mouse and 424 human proteins were
exclusively enriched by the OOPS method. While we cannot rule out false-positives, more
than 55% of the proteins of both subsets matched to EuRBPDB-listed annotations (not
shown). These findings suggested that genuine RBPs are found even outside of the
intersecting set of OOPs and RNA-IC identified proteins and that the definition of RBPomes
profits from employing different biochemical approaches.
T cell signaling proteins with unexpected RNA-binding function
Some of the identified RBPs of the core proteome including Stat1, Stat4 and Crip1 are not
expected to be associated with mRNA in cells. We therefore established an assay to confirm
RNA-binding of these candidates. To do so, GFP-tagged candidate proteins were
overexpressed in HEK293T cells, which were UV crosslinked, and extracts were used for
immunoprecipitations with GFP specific antibodies. By Western or North-Western blot
9 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
analyses with oligo(dT) probes we could verify pull-down of GFP-tagged proteins and the
association with mRNA for the RBPs Roquin-1 and Rbms1, but also for the lactate
dehydrogenase (Ldha) protein, a metabolic enzyme with known ability to also bind RNA 23
(Fig. 5a). Via this approach, the determined RNA association of Stat1, Stat4 and Crip1 by
global methods was indeed confirmed (Fig. 5b), but appeared less pronounced as
compared to prototypic RPBs, and similar with regard to Ldha (Fig 5a). This finding supports
a potential moonlighting function of these signaling proteins. As the identity of the associated
RNA species for these RBPs is unknown, we established a dual luciferase assay to
determine the impact of the different proteins on the expression of the renilla luciferase
transcript when they were tethered to an artificial 3’-UTR (Fig. 5c). We utilized the
λN/5xboxB system 41 and confirmed the expression of fusion proteins with a newly
established λN-specific antibody (Fig. 5d-e). Importantly, Stat1 and Stat4 repressed
luciferase function almost to the same extend as the known negative regulators Pat1b and
Roquin-1, or other known RBPs, such as Celf1, Rbms1 and Cpeb4 (Fig. 5f). λN-Crip1 and
λN-Vav1 expression did not exert a positive or negative influence, since their relative
luciferase expression appeared unchanged compared to cells transfected to only express
the λN polypeptide (Fig. 5f). These data suggest that the signaling proteins Stat1 and Stat4
that we defined here as part of the T helper cell RBPome not only have the capacity to bind
mRNA but can also exert RNA regulatory functions.
Analyzing higher order post-transcriptional regulation
We next devised an experimental strategy to uncover RBPs with the potential to antagonize
or cooperate with the Roquin-1 RBP in the repression of its target mRNAs by performing
‘BioID’ experiments (Fig. 6a). In this protein-centric, proximity-based labelling method we
expressed a Roquin-1 BirA* fusion protein to identify proteins that reside within a short
distance of approximately ~10 nm 42 in T cells (Fig. 6a). In this dataset we sought for
matches with the T cell RBPome (Fig. 4g) to identify proteins that shared the features,
‘RNA-binding’ and ‘Roquin-1 proximity’. We first verified that the mutated version of the biotin
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
ligase derived from E. coli (BirA*) which was N-terminally fused to Roquin-1 or GFP was
able to biotinylate residues in Roquin or other cellular proteins (Fig. 6b) but did not interfere
with the ability of Roquin-1 to downregulate Icos (Fig. 6c). Doxycycline-induced BirA*-
Roquin-1 compared to BirA*-GFP expression in CD4+ T cells significantly enriched biotin
labelling of 64 proteins (Supplementary Table 2), including Roquin-1 (Rc3h1) itself or
Roquin-2 (Rc3h2) (Fig. 6d) as well as previously identified Roquin-1 interactors and
downstream effectors, such as Ddx6 and Edc4 26, components of the Ccr4/Not complex 43, 44
and Nufip2 19 (Fig. 6d). More than half of all proteins in proximity to Roquin-1 were also part
of the defined RBPome (Fig. 6e-f). Increasing this BioID list with additional proteins that we
determined in proximity to Roquin-1 when establishing and validating the BioID method in
fibroblasts (Supplementary Fig. 6a-d), we arrived at 143 proteins (Supplementary Table
2) of which 96 (67%) were part of the RBPome (Supplementary Fig. 6e and
Supplementary Fig. 7a). Cloning of 46 candidate genes in the context of N-terminal GFP
fusions (Supplementary Fig. 8a) and generating retroviruses to overexpress candidate
proteins we analyzed their effect on endogenous Roquin-1 targets (Supplementary Fig.
8b). CD4+ T cells were used from mice with Rc3h1fl/fl;Rc3h2fl/fl;rtTA alleles in combination
with (iDKO) or without the CD4Cre-ERT2 allele (WT) allowing induced inactivation of
Roquin-1 and -2 by 4’-OH-tamoxifen treatment. Reflecting deletion, the Roquin-1 targets
Icos, Ox40, Ctla4, IκBNS and Regnase-1 became strongly derepressed in induced double-
knockout (iDKO) T cells (Supplementary Fig. 8c). This elevated expression was corrected
to wildtype levels in iDKO T cells that were retrovirally transduced and doxycycline-induced
to express GFP-Roquin-1 (Supplementary Fig. 8c). The target expression in WT T cells
was only moderately reduced through overexpression of GFP-Roquin-1 (Supplementary
Fig. 8c). For the majority of the 46 candidate genes induced expression in WT and iDKO
CD4+ T cells did not alter expression of the five analyzed Roquin-1 targets, exemplified here
by the results obtained for Vav1 (Fig. 7a-b). Interestingly, we identified a new function for
Rbms1 (transcript variant 2), specifically upregulating Ctla-4 (Fig. 7c-d). Furthermore, we
demonstrated that Cpeb4 strongly upregulates Ox40 and, most strikingly, in the same cells 11 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Cpeb4 repressed Ctla-4 levels (Fig. 7e-f). While these findings are new and noteworthy,
they occurred in a Roquin-1 independent manner. In contrast to these effects, we
determined a higher order regulation of Icos by Celf1. Induced Celf1 expression in wildtype T
cells clearly upregulated Icos, Ctla-4 and Ox40 expression and this function was obliterated
in Roquin-1-deficient iDKO cells (Fig. 7g-h). Importantly, this effect could not be explained
by Celf1-mediated repression of Roquin-1 on the protein or mRNA level (Supplementary
Fig. 8d-e). Together these findings demonstrate responsiveness of Roquin targets to many
other RBPs and a higher order, Roquin-dependent regulation of transcripts encoding for the
costimulatory receptors Icos, Ox40 and Ctla-4 by Celf1.
Discussion
The work on post-transcriptional gene regulation in T helper cells has focused on some
miRNAs and several RNA-binding proteins, and few reports described m6A RNA
methylation in this cell type. Although arriving at a more or less detailed understanding of
individual molecular relationships and regulatory circuits, this isolated knowledge assembles
into a very incomplete picture. Creating an atlas of the human and mouse T helper cell
RBPomes has now opened the stage, allowing to work towards understanding connections,
complexity and principles of post-transcriptional regulatory networks in these cells.
RNA-IC and OOPS are two complementary methods to define RBPs on a global scale.
While RNA-IC specifically queries for proteins bound to polyadenylated RNAs, OOPS
captures the RNA bound proteome in its whole. Applying both methods to T helper cells of
two different organisms allowed us to cross-validate the results from both methods and
solidified our description of the core mouse and human T helper cell RBPome. While the
vast majority of RNA-IC-identified CD4+ T cell RBPs were previously known RNA binders,
OOPS typically repeated and profoundly expanded these results (Supplementary Fig. 5),
with the possible caveat of identifying false-positive proteins. Contradicting this possibility at
least in part, more than half of OOPS identified proteins that were exclusively found in
mouse or man were EuRBPDB-listed.
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Strikingly, the signaling proteins Stat1 and Stat4 were identified by mouse RNA-IC and
human OOPS and were just below the cut-off in the mouse OOPS dataset, and we could
support their RBP function by additional RNA-binding assays. Undoubtedly, the defined
human and mouse T cell RBPomes contain many more unusual RBPs that would warrant
further investigations. We assume that even interactions of proteins without prototypic RBDs,
like the Stat proteins, with RNA will have consequences for both binding partners. As the
identity of the interacting mRNA(s) is unknown, we could only speculate about the post-
transcriptional impact. Nevertheless, Stat1 and Stat4 showed regulatory capacity in our
tethering assays. Intriguingly, RNA binding may also impact the function of Stat proteins as
transcription factors. Supporting this notion, early results found Stat1 bound to the non-
coding, polyadenylated RNA ‘TSU’, derived from a trophoblast cDNA library, and
translocation of Stat1 into the nucleus was reduced after TSU RNA microinjection into HeLa
cells 45, 46 .
Many 3’-UTRs, which effectively instruct post-transcriptional control, exhibit little sequence
conservation between species, and the exact modules which determine regulation are not
known. This for example is true for the Icos mRNA 19, 26. On the side of the trans-acting
factors, we find high similarity between the RBPomes of T helper cells of mouse and human
origin, actually reflecting the general overlap of so far determined RBPomes from many cell
lines of these species. We interpret this evolutionary conservation as holding ready similar
sets of RBPs in T helper cells across species, which are then able to work together on
composite cis-elements to reach comparable regulation of 3’-UTRs in the different
organisms, despite sequence variability and differences in the composition of elements.
We not only define the first RBPomes of human and mouse T helper cells. We also provide
potential avenues of how to make use of this information. Screening a set of candidates from
the T cell RBPome for effects on Roquin targets, our findings support a concept in which
post-transcriptional targets are separated into “regulons” 47. These regulons comprise
coregulated mRNA subsets responding to the same inputs and often functioning in the same
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
biological process. Therefore, complex and differential binding of targets by RBPs of the
cellular RBPome specifies the possible regulons and differential functions of the cell. Roquin
cooperated, coregulated, or antagonized in the regulation of Icos with Regnase-1, m6A
methylation and miRNA functions. Moreover, our screening approach added Rbms1 as
Roquin-independent and Celf1 as Roquin-dependent coregulator in the Icos containing
regulon. It also revealed that different targets responded very differently to the expression of
specific coregulators, as for example Ctla4 and Ox40 were inhibited or induced by the same
RBP, Cpeb4, respectively. Together these data indicated an unexpected wealth of possible
inputs from the T helper cell RBPome and suggested highly variable and combinatorial
mRNP compositions in higher order post-transcriptional gene regulation of individual targets.
To solve the seemingly simple question which RBPs regulate which mRNA in T helper cells,
we will require further knowledge about individual contributions, binding sites and composite
cis-elements, temporal and interdependent occupancies, interactions among RBPs and with
downstream effector molecules. In this endeavor global protein and RNA centric approaches
make fundamental contributions.
Acknowledgements
We thank Hemalatha Mutiah (Ludwig-Maximilians-Universität Munich) for screening
hybridoma supernatants and also Claudia Keplinger (Helmholtz Center Munich) for excellent
technical support. For the provision of mouse lines we would like to thank Marc Schmidt-
Supprian (Rc3h1), Wolfgang Wurst (Rc3h2), Robert Blelloch (Dgcr8flox), Mingui Fu
(Zc3h12aflox) and Thorsten Buch (Cd4-Cre-ERT2). The work was supported by the German
Research Foundation grants SPP-1935 (to VH), SFB-1054 projects A03 and Z02 as well as
HE3359/7-1 and HE3359/8-1 to V.H. as well as grants from the Wilhelm Sander, Fritz
Thyssen, Else Kro ner-Fresenius and Deusche Krebshilfe foundations to V.H.. D.B. was
supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
under Emmy Noether Programme BA 5132/1-1 and BA 5132/1‐2 (252623821), as well as
SFB 1054 project B12 (210592381). D.B. is a member of Germany's Excellence Strategy
EXC2151 577 (390873048).
Author contribution
EG and VH conceived the idea for the project and together with MW and MM supervised the
experimental work. KPH and VH wrote the manuscript with contributions from EG, MW and
AR. KPH performed OOPS, BioID and T cell transductions with help from GB, KD, SMH, JM
and CC. CG conducted RNA-IC experiments and RNA-binding assays. AR and SMH
performed mass spectrometry and AR analyzed RBPome data. SMH and KPH analyzed the
BioID data. MX contributed the tethering assays and EW analyzed Icos regulation for which
AG, RF, TI-K and DB provided unpublished reagents.
Data availability statement
The data sets generated and/or analyzed during the current study are available from the
corresponding authors on reasonable request.
Methods
Isolation, in vitro cultivation and transduction of primary CD4+ T cells
For in vitro cultivation of primary murine CD4+ T cells, mice were sacrificed and spleen as
well as cervical, axillary, brachial, inguinal and mesenteric lymph nodes were dissected and
pooled. Organs were squished and passed through a 100 μm filter under rinsing with T cell
isolation buffer (PBS supplemented with 2% FCS and 1 mM EDTA). Erythrocytes were
eliminated by incubating cells with TAC-lysis buffer (13 mM Tris, 140 mM NH4Cl, pH 7.2) for
5 min at room temperature. CD4+ T cells were isolated by negative selection using
EasySep™ Mouse CD4+ T cell isolation Kit (STEMCELL) according to manufacturer’s
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
protocol. Purified CD4+ T cells were cultured in DMEM (Invitrogen) T cell culture medium
supplemented with 10% FCS (Gibco), 1% Pen-Strep (Thermo Fisher), 1% HEPES-Buffer
(Invitrogen), 1% non-essential amino acids (NEAA; Invitrogen) and 1 mM β-mercaptoethanol
(Invitrogen). For activation and differentiation under TH1 conditions the T cells were
stimulated with α-CD3 (0.5 µg/mL; cl. 2C11, in house production), α-CD28 (2.5 µg/mL; cl.
37.5N, in house production), 10 μg/mL α-IL-4 (cl. 11B11, in house production) and 10 ng/mL
IL-12 (BD Pharmingen) and cultured on goat α-hamster IgG (MP Biochemicals) pre-coated
6-or 12-well culture plates for 40-48h at an initial cell density of 5 or 1.5×106 cells/mL,
respectively. The cells were then resuspended and cultured in medium supplemented with
200 IE/mL recombinant hIL-2 (Novartis) in a 10% CO2 incubator and expanded for 2-4 days,
as indicated. Subsequently cells were fed with fresh IL-2-containing medium every 24h and
cultured at a density of 0.5-1×106 cells/mL. For in vitro deletion of floxed alleles of Rc3h1fl/fl
;Rc3h2fl/fl ;CD4Cre-ERT2;rtTA3 (but with no effect on the Cre-deficient WT control Rc3h1fl/fl
;Rc3h2fl/fl ;rtTA3) CD4+ T cells were treated with 1 µM 4´OH-tamoxifen (Sigma) for 24h prior
to activation at a cell density of 0,5-1×106 cells/mL. We performed retroviral transduction 40h
after the start of anti-CD3/CD28 activation, T cells were transduced with retroviral particles
using spin-inoculation (1h, 18 °C, 850 g), and after 4-6h co-incubation of T cells and virus,
virus particles were removed and T cells resuspended in T cell medium supplemented with
IL-2 as described before. To induce expression of pRetro-Xtight-GFP constructs in rtTA
expressing T cells, the transduced cells were cultured for 16h in the presence of doxycycline
(1 µg/mL) prior to flow cytometry analysis of expression of targets in transduced GFP+ cells.
In vivo deletion of loxP-flanked alleles and in vitro culture of CD4+ T cells
For in vitro culture analysis, deletion of Roquin (Rc3h1fl/fl;Rc3h2fl/fl;CD4-Cre-ERT2) 48,
Regnase-1 (Zc3h12afl/fl;CD4-Cre-ERT2) 49, Wtap (Wtapfl/fl;CD4-Cre-ERT2) 33, and Dgcr8
(Dgcr8fl/fl;CD4-Cre-ERT2) 34 encoding alleles in Cd4-cre-ERT2 mice was induced in vivo by
oral transfer of 5 mg tamoxifen (Sigma) in corn oil. Two doses of tamoxifen each day were
given on two consecutive days (total of 20 mg tamoxifen per mouse). Mice with the genotype
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
CD4-Cre-ERT2 (without floxed alleles) were used for wild-type controls. Mice were scarified
three days after the last gavage and total CD4+ T cells were isolated using the EasySepTM
Mouse T Cell Isolation Kit (Stem Cell) and activated under TH1 conditions as described
above.
Flow cytometry
Following in vivo deletion and CD4+ T cell isolation (above) cells were activated and cultured
under TH1 conditions. Cells were obtained daily for FACS analysis. The single cell
suspensions were stained with fixable violet dead cell stain (Thermo Fisher) for 20 min at
4°C. For the detection of surface proteins, cells were stained with the appropriate antibodies
in FACS buffer for 20 min at 4°C. After staining, cells were acquired on a FACS Canto II (3-
laser). The data were further processed with the software FlowJo 10 software (BD
Bioscience). The following antibodies were used: anti-CD4 (cl. GK1.5), anti-CD44 (cl. IM7),
anti-CD62L (cl. MEL-14), anti-Icos (cl. C398.4A), anti-Ox40 (cl. OX-86), all from eBioscience,
anti-CD25 (cl. PC61, Biolegend).
Effects of doxycycline-induced expression of 46 GFP-GOI fusion protein in 2x106 wild-type
and Roquin iDKO CD4+ T cells were analyzed on day 6 after isolation (compare Suppl. Fig.
8b). First, proteins were treated with a fixable blue dead cell stain (Invitrogen) and after
washing, stained in three panels to interrogate the surface expression of Icos and Ox40
(Icos-PE, clone 7E.17G9/Ox40-APC, clone OX-86; both eBioscience) and to intracellularly
measure Ctla4, IκBNS (Ctla4-PE, UC10-4B9; eBioscience/cl. 4C1 rat monoclonal; in house
production) as well as Regnase-1 (cl. 15D11 rat monoclonal; in house production). For
intracellular staining cells were fixed in 2% formaldehyde for 15 min at RT, permeabilized by
washing in Saponin buffer and stained with the appropriate antibodies for 1h at 4 °C. After
washing, anti-rat-AF647 antibody (cl. poly4054; Biolegend) was added for 30 min. After
additional rounds of Saponin- and FACS buffer washing, acquisition was performed using a
LSR Fortessa.
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Isolation and differentiation of effector and regulatory T cells for RNA-IC
Naïve CD4+ T cells were isolated by using Dyna- and Detachabeads (11445D and 12406D,
Invitrogen) from spleens and mesenteric lymph nodes of 8-12 week old C57BL/6J mice. For
iTreg culture, cells were additionally selected for CD62L+ with anti-CD62L (clone: Mel14)
coated beads. All cells were then activated with plate-bound anti-CD3 (using first anti-
hamster, 55397, Novartis, then anti-CD3 in solution: clone: 2C11H: 0.1 µg/mL) and soluble
anti-CD28 (clone: 37N: 1 µg/mL) and cultured in RPMI medium (supplemented with 10%
(vol/vol) FCS, β-mercaptoethanol (0.05 mM, Gibco), penicillin-streptomycin (100 U/ml,
Gibco), Sodium Pyruvate (1 mM, Lonza), Non-Essential Amino Acids (1x, Gibco), MEM
Vitamin Solution (1x, Gibco), Glutamax (1x, Gibco) and HEPES pH 7.2 (10 mM, Gibco)). For
iTreg cell differentiation we additionally added the following cytokines and blocking
antibodies: rmIL-2 and rmTGF-β (both: R&D Systems, 5 ng/ml), anti-IL-4 (clone: 11B11, 10
µg/ml) and anti-IFNγ (clone: Xmg-121, 10 µg/ml). All antibodies were obtained in
collaboration with and from Regina Feederle (Helmholtz Center Munich). After differentiation
for 36-48 h cells were expanded for 2-3 days. iTreg cells were cultured in RPMI and 2000
units Proleukin S (02238131, MP Biomedicals) and Teff cells with 200 U Proleukin S. We only
used iTreg cells for experiments if samples achieved at least 80% Foxp3 positive cells
(00552300, Foxp3 Staining Kit, BD Bioscience).
EL-4 T cells were cultured in the same medium as primary T cells. HEK293T cells were
cultured in DMEM (supplemented with 10% (vol/vol) FCS, penicillin-streptomycin (100 U/ml,
Gibco) and Hepes, pH 7.2 (10 mM, Gibco)).
RBPome capture
6 For RBPome capture for mass spectrometry, 20 x 10 Teff or iTreg cells were either lysed
directly (nonirradiated, control) in 1 ml lysis buffer from the µMACS mRNA Isolation Kit (130-
075-201, Miltenyi) or suspended in 1 ml PBS, dispensed on a 10 cm dish and UV irradiated
at 0.2 J/cm2 at 254 nm for 1 min, washed with PBS, pelleted and subsequently lysed (UV 18 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
irradiated) and mRNA was isolated from both samples with the µMACS mRNA Isolation Kit
according to the manufacturer’s instructions. RNAs and crosslinked proteins were eluted
6 with 70° C RNase-free H2O. For RBPome capture for Western blot analysis, 400 x 10 EL-4
T cells were either lysed directly in 8 ml lysis buffer (nonirradiated, control) or suspended in
16 ml PBS, dispensed on sixteen 10 cm dishes and UV irradiated at 0.2 J/cm2 at 254 nm for
1 min, washed with PBS, pelleted and subsequently lysed in 8 ml lysis buffer (UV irradiated)
and then mRNA was isolated from both samples with the µMACS mRNA Isolation Kit using
500 µl oligo(dT) beads per sample. Each sample was split and run over two M columns
(130-042-801, Miltenyi) and each column was eluted with two times 100 µl RNase-free H2O.
The eluate was concentrated in Amicon centrifugal filter units (UFC501008, Merck) to a final
volume of ∼25 µl. 8 µl Lämmli buffer (4x) with 10% (vol/vol) β-mercaptoethanol was added
and the sample boiled for 5 min at 95° C. For protein analysis, samples were either flash-
frozen in liquid nitrogen for MS analysis or Lämmli loading dye was added for subsequent
analysis by Western blotting or silver staining.
OOPS
20-30 x 106 mouse CD4+ T cells were isolated as described above and activated without
bias. Cells were washed in PBS once and resuspended in 1 ml of cold PBS and transferred
into one well of a six well dish. Floating on ice, the cells in the open 6 well plate were UV
irradiated once at 0.4 J/cm2 and twice at 0.2 J/cm2 at 254 nm with shaking in-between
sessions. After irradiation the 1ml of cell suspension was transferred into a FACS tube and
the well was washed with another 1 ml of cold PBS which was also added to the FACS tube.
After centrifugation the cell pellet was completely dissolved in 1 ml of Trizol (Sigmal). The
remainder of the procedure was performed strictly according to 39 with the exception that we
broke up and regenerated interphases in five successive rounds of phase partitioning, rather
than three.
Culture preparation of human CD4+ T cell blasts 19 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Blood (120 ml) was collected by venepuncture from two times four donors for RNA-IC and
OOPS. Peripheral blood mononuclear cells (PBMCs) were isolated by standard Ficoll
gradient (PancollR) centrifugation and CD4+ T cells were isolated from 1x108 cells using
CD4+ Microbeads (Miltenyi) to arrive at 2-3x107 cells. These were resuspended at 2x106/ml
in T cell medium ((AIM-V (Invitrogen), 10% heat-inactivated human serum, 2mM L-
glutamine, 10 mM HEPES and 1,25 µg/ml Fungizone), supplemented with 500 ng/ml PHA
(Murex) and 100 IU IL-2/ml (Novartis). Cells were dispensed in 24-well plates at 2 ml per
well and on day 3 old medium was replaced by fresh T cell medium and cell were
harvested and counted at day 4,5. For generating iTreg, naïve CD4+ T cells were isolated
from PBMCs using Microbeads (Naive CD4+ T Cell Isolation Kit II, Miltenyi) and
resuspended at 1x106 cells/ml in T cell medium supplemented with 500 ng/ml PHA (Murex),
500 IU IL-2/ml (Novartis), 500 nM Rapamycin (Selleckchem), and 5 ng/ml TGF-ß1
(Miltenyi). Cells were cultured in 24-well plates at 2 ml per well together with 1x106
irradiated (40 Gy) PBMCs derived from three donors. On day 3, old medium was replaced
by fresh T cell medium including supplements and the cells harvested and counted at day
5.
SDS-PAGE, Western blotting and silver staining
All protein visualization procedures were performed according to standard protocols. For
silver staining we used the SilverQuest kit (LC6070, Invitrogen). Antibodies used were anti-
GFP, (1:10, clone: 3E5-111, in house), anti-Ptbp1, (1:1000, 8776, Cell Signaling), anti-β-
tubulin, (1:1000, 86298, Cell Signaling). Proteins were visualized by staining with anti-rat
(1:3000, 7077, Cell Signaling) or anti-mouse (1:3000, 7076, Cell Signaling) secondary
antibodies conjugated to HRP.
Sample preparation for mass spectrometry
For RNA-capture, eluates were incubated with 10 µg/ml RNase A in 100 mM Tris, 50 mM
NaCl, 1 mM EDTA at 37°C for 30 min. RNase-treated eluates were acetone precipitated and
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
resuspended in denaturation buffer (6 M urea, 2 M thiourea, 10 mM Hepes, pH 8), reduced
with 1 mM DTT and alkylated with 5.5 mM IAA. Samples were diluted 1:5 with 62.5 mM Tris,
pH 8.1 and proteins digested with 0.5 µg Lys-C and 0.5 µg Trypsin at room temperature
overnight. The resulting peptides were desalted using stage-tips containing three layers of
C18 material (Empore).
For OOPS experiments, 100 µl of lysis buffer (Preomics, iST kit) were added and samples
incubated at 100°C for 10 min at 1,400 rpm. Samples were sonicated for 15 cycles (30s
on/30s off) on a bioruptor (Diagenode). Protein concentration was determined using the BCA
assay and about 30 µg of proteins were digested. To this end, trypsin and Lys-C were added
in a 1:100 ratio, samples diluted with lysis buffer to contain at least 50 µl of volume and
incubated overnight at 37°C. To 50 µl of sample, 250 ul Isopropanol/1% TFA were added
and samples vortexed for 15s. Samples were transferred on SDB-RPS (Empore) stagetips
(3 layers), washed twice with 100 µl Isopropanol/1% TFA and twice with 100 µl 0.2% TFA.
Peptides were eluted with 80 µl of 2% ammonia/80% acetonitrile, evaporated on a
centrifugal evaporator and resuspended with 10 µl of buffer A* (2% ACN, 0.1% TFA).
LC-MS/MS analysis
Peptides were separated on a reverse phase column (50 cm length, 75 μm inner diameter)
packed in-house with ReproSil-Pur C18-AQ 1.9 µm resin (Dr. Maisch GmbH). Reverse-
phase chromatography was performed with an EASY-nLC 1000 ultra-high pressure system,
coupled to a Q-Exactive HF Mass Spectrometer (Thermo Scientific) for mouse RNA-capture
experiments or a Q-Exactive HF-X Mass Spectrometer (Thermo Scientific) for human RNA-
capture, OOPS experiments and single-shot proteomes. Peptides were loaded with buffer A
(0.1% (v/v) formic acid) and eluted with a nonlinear 120-min (100-min gradient for human
RNA-capture and OOPS experiments) gradient of 5–60% buffer B (0.1% (v/v) formic acid,
80% (v/v) acetonitrile) at a flow rate of 250 nl/min (300 nl/min for human RNA-capture and
OOPS). After each gradient, the column was washed with 95% buffer B and re-equilibrated
with buffer A. Column temperature was kept at 60° C by an in-house designed oven with a
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Peltier element and operational parameters were monitored in real time by the SprayQc
software. MS data were acquired using a data-dependent top 15 (top 12 for human RNA-
capture and OOPS experiments) method in positive mode. Target value for the full scan MS
spectra was 3 × 106 charges in the 300−1,650 m/z range with a maximum injection time of
20 ms and a resolution of 60,000. The precursor isolation windows was set to 1.4 m/z and
capillary temperature was 250°C. Precursors were fragmented by higher-energy collisional
dissociation (HCD) with a normalized collision energy (NCE) of 27. MS/MS scans were
acquired at a resolution of 15,000 with an ion target value of 1 × 105, a maximum injection
time of 120 ms (60 ms for human RNA-capture and OOPS experiments). Repeated
sequencing of peptides was minimized by a dynamic exclusion time of 20 s (30 ms for
human RNA-capture and OOPS).
Raw data processing
MS raw files were analyzed by the MaxQuant software50 (version 1.5.1.6 for RNA-capture
files and version 1.5.6.7 for OOPS files) and peak lists were searched against the mouse or
human Uniprot FASTA database, respectively, and a common contaminants database (247
entries) by the Andromeda search engine51. Cysteine carbamidomethylation was set as fixed
modification, methionine oxidation and N-terminal protein acetylation as variable
modifications. False discovery rate was 1% for both proteins and peptides (minimum length
of 7 amino acids). The maximum number of missed cleavages allowed was 2. Maximal
allowed precursor mass deviation for peptide identification was 4.5 ppm after time-
dependent mass calibration and maximal fragment mass deviation was 20 ppm. Relative
quantification was performed using the MaxLFQ algorithm52. “Match between runs” was
activated with a retention time alignment window of 20 min and a match time window of 0.5
min for RNA-capture experiments, while matching between runs was disabled for OOPS
experiments. The minimum ratio count was set to 2 for label-free quantification.
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Data analysis
Statistical analysis of MS data was performed using Perseus (version 1.6.0.28). Human
RNA-capture, mouse RNA-capture, human OOPS and mouse OOPS data was processed
separately. For all experiments, MaxQuant output tables were filtered to remove protein
groups matching the reverse database, contaminants or proteins only observed with
modified peptides. Next, protein groups were filtered to have at least two valid values in
either the crosslinked or control triplicate. LFQ intensities were logarithmized (base 2) and
missing values were imputed from a normal distribution with a downshift of 1.8 standard
deviations and a width of 0.2 (0.25 for OOPS data). For RNA-capture experiments, a
Student’s T-test was performed to find proteins significantly enriched in the crosslinked
sample over the non-crosslinked control (false-discovery rate (FDR) < 0.05). As many
proteins were identified specifically in the crosslinked sample at intensities too low to find
significant differences compared to the imputed values, we additionally considered proteins
only identified in two or three replicates of the crosslinked sample, but never in the non-
crosslinked control, as RNA-binding proteins. For OOPS experiments, proteins significantly
enriched in a Student’s T-tests of the organic phase after RNase-treatment over the same
sample of the non-crosslinked control (FDR < 0.05) were considered RBPs.
GO term enrichment analysis was performed using the clusterProfiler package in R (version
4.1.0) as described in the original publication53. The mouse or human proteomes served as
background for the respective enrichment analysis. Relative abundance of proteins in the
single-shot proteome (Fig. 4a-b) was determined by calculating the logarithm (base 2) of the
ratio of the LFQ intensity and the number of theoretical peptides. Relative abundance of
proteins significant in the mouse OOPS and RNA-IC is shown in the single-shot proteome of
mouse CD4+ T cells measured with the OOPS samples. Relative abundance of proteins
significant in the human OOPS and RNA-IC experiments is shown in the single-shot
proteome of human CD4+ T cells measured with the OOPS samples or RNA-IC samples,
respectively. For the 4-way Venn comparison to find RBPs identified in more than one RNA-
IC or OOPS experiment, human gene names were converted into their homologous mouse
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
counterparts. Protein group entries containing more than one isoform were expanded for the
comparison and subsequently collapsed into one entry again for calculation of the Venn
diagram. Multiple gene name entries for different unambiguously identified isoforms were
collapsed into the major isoform. This affected six protein groups in the human RNA-IC
dataset and three protein groups in the mouse RNA-IC dataset.
Intrinsically disordered regions were retrieved from the Disorder Atlas54. We referred to the
LCR-eXXXplorer55 to obtain low complexity regions in proteins.
Plasmid construction
To generate a vector that expresses N-terminally GFP-tagged proteins, we amplified the
respective genes from cDNA of Teff or TFoxp3+ cells, added HindIII and KpnI restriction sites in
front of the start codon and cloned them into the pCR™8/GW/TOPO® vector. We then used
HindIII and KpnI to insert a GFP sequence where we removed the bases for the stop codon.
The respective sequences were subsequently transferred to the expression vector pMSCV
via the gateway cloning technology. Only GFP-Roquin-1 (a kind gift of Vigo Heissmeyer)
was expressed from the vector pDEST14. For oligonucleotide sequences see
Supplementary Table 3.
Validation of RNA binding ability
HEK293T cells were transfected by calcium phosphate transfection with plasmids
expressing the respective proteins with an N-terminal GFP-tag or GFP alone. After three
days, cells were washed with PBS on plates, UV crosslinked (CL) as before or directly
scraped from the plates (nCL). Cell lysates were generated by flash-freezing pellets in liquid
nitrogen and incubation in NP-40 lysis buffer (150 mM NaCl, 1% NP-40, 50 mM Tris-HCl, pH
7.4, 5 mM EDTA, 1 mM DTT, 1 mM PMSF and protease inhibitor mixture (Complete,
Roche)). After lysis, extracts were cleared by centrifugation at 17000 g for 15 min at 4° C.
We then determined protein concentration via the BCA method and used 2-10 mg of protein
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
for the subsequent GFP immunoprecipitation, depending on transfection efficiency and
expected RNA-binding capacity. We pre-coupled 200 µl Protein-G beads (10004D,
Dynabeads Protein G, Invitrogen) with 20 µg antibody (anti-GFP, clone: 3E5-111, in house)
in PBS (1 h, RT), washed beads in lysis buffer, added them to cell lysates and incubated
with rotation for 4 h at 4 °C. Beads were then washed three times with IP wash buffer (50
mM Tris-HCl, pH 7.5) with decreasing salt (500 mM, 350 mM, 150 mM, 50 mM NaCl) and
SDS (0.05%, 0.035%, 0.015%, 0.005%) concentrations. Proteins and crosslinked RNAs
were eluted with 50 mM glycine, pH 2.2 at 70° C for 5 min. Lämmli buffer (4x) was added
and samples were divided for mRNA and protein detection and separated via SDS gel
electrophoresis (6% SDS gels for detection of mRNA samples and 9% gels to verify
immunoprecipitation efficiency). For RNA detection we blotted onto Nitrocellulose
membranes and for protein detection on PVDF membranes. After transfer, the Nitrocellulose
membrane was prehybridized with Church buffer (0.36 M Na2HPO4, 0.14 M NaH2PO4, 1 mM
EDTA, 7% SDS) for 30 min and then incubated for 4 h with Church buffer containing 40 nM
3´-and 5´-Biotin labeled oligo(dT)20 probe to anneal to the poly-A tail of the bound mRNA.
The membrane was washed twice with 1 x SSC, 0.5% SDS and twice with 0.5 x SSC, 0.5%
SDS. Bound mRNA was detected with the Chemiluminescent Nucleic Acid Detection Kit
Module (89880, Thermo Fisher) according to the manufacturer´s instructions.
Real-Time PCR
Total RNA of input was purified from lysates with Agencourt RNAClean XP Beads (A63987,
Beckman Coulter) according to the manufacturer´s instructions and eluted in nuclease-free
H2O. cDNA was synthesized from total input RNA and oligo(dT)-isolated RNA with the
QuantiTect Reverse Transcription Kit (205311, Qiagen). All qRT-PCRs were performed with
the SYBR green method. For Primer sequences see Supplementary Table 3.
RNA isolation, reverse transcription and quantitative RT PCR as shown in Figure 8d was
performed as published 56 using the universal probes systems (Roche). Primers for Rc3h1
(F: gagacagcaccttaccagca; R: gacaaagcgggacacacat; probe 22) and Hprt (F:
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Tgatagatccattcctatgactgtaga; R: aagacattctttccagttaaagttgag; probe 95) were efficiency
tested (both E=1.99).
Tethering assay
Hela cells were seeded in 24-well plates using 5×104 cells per well. Transfection was
performed the following day using Lipofectamine2000 (Invitrogen) and 300 ng of total
constructs. Each transfection consisted of 75 ng of luciferase reporter plasmid psiCHECK2
(Promega) or luciferase-5boxB plasmid psiCHECK2 -5boxB, 225 ng of pDEST12.2-λN fused
constructs. After 24 h, cells were harvested for luciferase activity assays using a Dual-
Luciferase Reporter Assay System (Promega). Renilla luciferase activity was normalized to
Firefly luciferase activity in each well to control for variation in transfection efficiency.
psiCHECK2 lacking boxB sites served as a negative control, and each transfection was
analysed in triplicates.
BioID
The proximity-dependent biotin identification assay was performed according to Roux (Roux
et al., 2012) with modifications. For each sample 2x107 MEF cells were grown on ten 15-cm
cell culture dishes for 24h before BirA*-Roquin-1 or BirA* expression was induced by
doxycycline treatment. For T cells, transduction with the same BirA*-fusions cloned into the
plasmid pRetroXtight was performed as described above and the same number of cells was
used for the experiment. Six hours after addition of doxyclycline, biotin was added for 16h to
arrive at an end concentration of 50 µM. Approximately 8x107 cells per sample were
trypsinized, washed twice with PBS and lysed in 5 ml lysis buffer (50 mM Tris-HCL, ph7,4;
500 mM NaCl, 0,2% SDS; 1x protease inhibitors (Roche), 20 mM DTT, 25 U/ml Benzonase)
for 30 min at 4 °C using an end-over-end mixer. After adding 500 µl of 20% Triton X-100 the
samples were sonicated for two sessions of 30 pulses at 30% duty cycle and output level 2,
using a Branson Sonifier 450 device. Keep on ice for two minutes in between sessions.
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Pipetting of 4,5 ml prechilled 50 mM Tris-HCL, pH 7,4 was followed by an additional round of
sonication. During centrifugation at 16500 x g for 10 min at 4 °C 500 µl streptavidin beads
(Invitrogen) or each sample was equilibrated in a 1:1 mixture of lysis buffer and 50 mM Tris-
HCL pH 7,4. After overnight binding on a rotator at 4 °C Streptavidin beads were stringently
washed using wash buffers 1, 2 and 3 (Roux et al, 2012) and prepared for mass
spectrometry by three additional washes with buffer 4 (1 mM EDTA, 20 mM NaCl, 50 mM
Tris-HCL pH 7,4). Proteins were eluted from streptavidin beads with 50 µl of biotin-saturated
1x sample buffer (50 mM Tris-HCL pH 6,8, 12% sucrose, 2% SDS, 20 mM DTT, 0,004%
Bromphenol blue, 3 mM Biotin) by incubation for 7 min at 98 °C. For identification and
quantification of proteins, samples were proteolysed by a modified filter aided sample
preparation 57 and eluted peptides were analysed by LC-MSMS on a QExactive HF mass
spectrometer (ThermoFisher Scientific) coupled directly to a UItimate 3000 RSLC nano-
HPLC (Dionex). Label-free quantification was based on peptide intensities from extracted ion
chromatograms and performed with the Progenesis QI software (Nonlinear Dynamics,
Waters). Raw files were imported and after alignment, filtering and normalization, all MSMS
spectra were exported and searched against the Swissprot mouse database (16772
sequences, Release 2016_02) using the Mascot search engine with 10 ppm peptide mass
tolerance and 0.02 Da fragment mass tolerance, one missed cleavage allowed, and
carbamidomethylation set as fixed modification, methionine oxidation and asparagine or
glutamine deamidation allowed as variable modifications. A Mascot-integrated decoy
database search calculated an average false discovery of < 1% when searches were
performed with a mascot percolator score cut-off of 13 and significance threshold of 0.05.
Peptide assignments were re-imported into the Progenesis QI software. For quantification,
only unique peptides of an identified protein were included, and the total cumulative
normalized abundance was calculated by summing the abundances of all peptides allocated
to the respective protein. A t-test implemented in the Progenesis QI software comparing the
normalized abundances of the individual proteins between groups was calculated and
corrected for multiple testing resulting in q values (FDR adjusted p values) given in
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplemental Table 2. Only proteins identified by two or more peptides were included in
the list of Roquin-1 proximal proteins.
Antibodies
To generate monoclonal antibodies against pan-Ythdf proteins or λN-peptide, Wistar rats
were immunized with purified GST-tagged full-length mouse Ythdf3 protein or ovalbumin-
coupled peptide against λN (MNARTRRRERRAEKQWKAAN) using standard procedures as
described 58. The hybridoma cells of Ythdf- or λN-reactive supernatants were cloned at least
twice by limiting dilution. Experiments in this study were performed with anti-pan-Ythdf clone
DF3 17F2 (rat IgG2a/κ) and anti-λN clone LAN 4F10 (rat IgG2b/κ).
References
1. Hoefig, K.P. & Heissmeyer, V. Degradation of oligouridylated histone mRNAs: see UUUUU and goodbye. Wiley interdisciplinary reviews. RNA 5, 577-589 (2014).
2. Hoefig, K.P. & Heissmeyer, V. Posttranscriptional regulation of T helper cell fate decisions. The Journal of cell biology (2018).
3. Salerno, F., Turner, M. & Wolkers, M.C. Dynamic Post-Transcriptional Events Governing CD8(+) T Cell Homeostasis and Effector Function. Trends in immunology 41, 240-254 (2020).
4. Shulman, Z. & Stern-Ginossar, N. The RNA modification N(6)-methyladenosine as a novel regulator of the immune system. Nature immunology 21, 501-512 (2020).
5. Turner, M. & Hodson, D.J. An emerging role of RNA-binding proteins as multifunctional regulators of lymphocyte development and function. Adv Immunol 115, 161-185 (2012).
6. Chen, J. et al. Posttranscriptional gene regulation of IL-17 by the RNA-binding protein HuR is required for initiation of experimental autoimmune encephalomyelitis. Journal of immunology 191, 5441-5450 (2013).
7. Moore, M.J. et al. ZFP36 RNA-binding proteins restrain T cell activation and anti-viral immunity. eLife 7 (2018).
28 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
8. Stellato, C. et al. Coordinate regulation of GATA-3 and Th2 cytokine gene expression by the RNA-binding protein HuR. Journal of immunology 187, 441-449 (2011).
9. Vogel, K.U., Bell, L.S., Galloway, A., Ahlfors, H. & Turner, M. The RNA-Binding Proteins Zfp36l1 and Zfp36l2 Enforce the Thymic beta-Selection Checkpoint by Limiting DNA Damage Response Signaling and Cell Cycle Progression. Journal of immunology 197, 2673-2685 (2016).
10. Jeltsch, K.M. et al. Cleavage of roquin and regnase-1 by the paracaspase MALT1 releases their cooperatively repressed targets to promote T(H)17 differentiation. Nature immunology 15, 1079-1089 (2014).
11. Vogel, K.U. et al. Roquin paralogs 1 and 2 redundantly repress the Icos and Ox40 costimulator mRNAs and control follicular helper T cell differentiation. Immunity 38, 655-668 (2013).
12. Minagawa, K. et al. Posttranscriptional modulation of cytokine production in T cells for the regulation of excessive inflammation by TFL. Journal of immunology 192, 1512-1524 (2014).
13. Uehata, T. et al. Malt1-induced cleavage of regnase-1 in CD4(+) helper T cells regulates immune activation. Cell 153, 1036-1049 (2013).
14. Baumjohann, D. & Ansel, K.M. MicroRNA-mediated regulation of T helper cell differentiation and plasticity. Nature reviews. Immunology 13, 666-678 (2013).
15. Akira, S. Regnase-1, a ribonuclease involved in the regulation of immune responses. Cold Spring Harb Symp Quant Biol 78, 51-60 (2013).
16. Heissmeyer, V. & Vogel, K.U. Molecular control of Tfh-cell differentiation by Roquin family proteins. Immunological reviews 253, 273-289 (2013).
17. Hodson, D.J., Screen, M. & Turner, M. RNA-binding proteins in hematopoiesis and hematological malignancy. Blood 133, 2365-2373 (2019).
18. Turner, M., Galloway, A. & Vigorito, E. Noncoding RNA and its associated proteins as regulatory elements of the immune system. Nature immunology 15, 484-491 (2014).
19. Rehage, N. et al. Binding of NUFIP2 to Roquin promotes recognition and regulation of ICOS mRNA. Nature communications 9, 299 (2018).
20. Mukherjee, N. et al. Global target mRNA specification and regulation by the RNA- binding protein ZFP36. Genome biology 15, R12 (2014).
21. Masuda, K. et al. Arid5a controls IL-6 mRNA stability, which contributes to elevation of IL-6 level in vivo. Proceedings of the National Academy of Sciences of the United States of America 110, 9409-9414 (2013).
29 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
22. Gerstberger, S., Hafner, M. & Tuschl, T. A census of human RNA-binding proteins. Nature reviews. Genetics 15, 829-845 (2014).
23. Castello, A. et al. Insights into RNA biology from an atlas of mammalian mRNA- binding proteins. Cell 149, 1393-1406 (2012).
24. Hentze, M.W., Castello, A., Schwarzl, T. & Preiss, T. A brave new world of RNA- binding proteins. Nat Rev Mol Cell Biol 19, 327-341 (2018).
25. Liao, J.Y. et al. EuRBPDB: a comprehensive resource for annotation, functional and oncological investigation of eukaryotic RNA binding proteins (RBPs). Nucleic acids research 48, D307-D313 (2020).
26. Glasmacher, E. et al. Roquin binds inducible costimulator mRNA and effectors of mRNA decay to induce microRNA-independent post-transcriptional repression. Nature immunology 11, 725-733 (2010).
27. Pratama, A. et al. MicroRNA-146a regulates ICOS-ICOSL signalling to limit accumulation of T follicular helper cells and germinal centres. Nature communications 6, 6436 (2015).
28. Vinuesa, C.G. et al. A RING-type ubiquitin ligase family member required to repress follicular helper T cells and autoimmunity. Nature 435, 452-458 (2005).
29. Zhu, Y. et al. The E3 ligase VHL promotes follicular helper T cell differentiation via glycolytic-epigenetic control. The Journal of experimental medicine 216, 1664-1681 (2019).
30. Zaccara, S., Ries, R.J. & Jaffrey, S.R. Reading, writing and erasing mRNA methylation. Nat Rev Mol Cell Biol 20, 608-624 (2019).
31. Edupuganti, R.R. et al. N(6)-methyladenosine (m(6)A) recruits and repels proteins to regulate mRNA homeostasis. Nature structural & molecular biology 24, 870-878 (2017).
32. Liu, N. et al. N(6)-methyladenosine-dependent RNA structural switches regulate RNA-protein interactions. Nature 518, 560-564 (2015).
33. Weichmann, F. et al. Validation strategies for antibodies targeting modified ribonucleotides. RNA (2020).
34. Rao, P.K. et al. Loss of cardiac microRNA-mediated regulation leads to dilated cardiomyopathy and heart failure. Circ Res 105, 585-594 (2009).
35. Sledzinska, A. et al. TGF-beta signalling is required for CD4(+) T cell homeostasis but dispensable for regulatory T cell function. PLoS Biol 11, e1001674 (2013).
30 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
36. Bronevetsky, Y. et al. T cell activation induces proteasomal degradation of Argonaute and rapid remodeling of the microRNA repertoire. The Journal of experimental medicine 210, 417-432 (2013).
37. Bohjanen, P.R., Moua, M.L., Guo, L., Taye, A. & Vlasova-St Louis, I.A. Altered CELF1 binding to target transcripts in malignant T cells. RNA 21, 1757-1769 (2015).
38. Castello, A. et al. Comprehensive Identification of RNA-Binding Domains in Human Cells. Mol Cell 63, 696-710 (2016).
39. Queiroz, R.M.L. et al. Comprehensive identification of RNA-protein interactions in any organism using orthogonal organic phase separation (OOPS). Nat Biotechnol 37, 169-178 (2019).
40. Villanueva, E. et al. Efficient recovery of the RNA-bound proteome and protein- bound transcriptome using phase separation (OOPS). Nat Protoc (2020).
41. Amaya Ramirez, C.C., Hubbe, P., Mandel, N. & Bethune, J. 4EHP-independent repression of endogenous mRNAs by the RNA-binding protein GIGYF2. Nucleic acids research 46, 5792-5808 (2018).
42. Roux, K.J., Kim, D.I., Raida, M. & Burke, B. A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. The Journal of cell biology 196, 801-810 (2012).
43. Leppek, K. et al. Roquin promotes constitutive mRNA decay via a conserved class of stem-loop recognition motifs. Cell 153, 869-881 (2013).
44. Sgromo, A. et al. A CAF40-binding motif facilitates recruitment of the CCR4-NOT complex to mRNAs targeted by Drosophila Roquin. Nature communications 8, 14307 (2017).
45. Peyman, J.A. Repression of major histocompatibility complex genes by a human trophoblast ribonucleic acid. Biol Reprod 60, 23-31 (1999).
46. Peyman, J.A. Mammalian expression cloning of two human trophoblast suppressors of major histocompatibility complex genes. Am J Reprod Immunol 45, 382-392 (2001).
47. Keene, J.D. RNA regulons: coordination of post-transcriptional events. Nature reviews. Genetics 8, 533-543 (2007).
48. Tavernier, S.J. et al. A human immune dysregulation syndrome characterized by severe hyperinflammation with a homozygous nonsense Roquin-1 mutation. Nature communications 10, 4779 (2019).
31 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
49. Li, Y. et al. Central role of myeloid MCPIP1 in protecting against LPS-induced inflammation and lung injury. Signal Transduct Target Ther 2, 17066 (2017).
50. Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol 26, 1367-1372 (2008).
51. Cox, J. et al. Andromeda: a peptide search engine integrated into the MaxQuant environment. J Proteome Res 10, 1794-1805 (2011).
52. Cox, J. et al. Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol Cell Proteomics 13, 2513-2526 (2014).
53. Yu, G., Wang, L.G., Han, Y. & He, Q.Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 16, 284-287 (2012).
54. Vincent, M. & Schnell, S. Disorder Atlas: Web-based software for the proteome- based interpretation of intrinsic disorder predictions. Comput Biol Chem 83, 107090 (2019).
55. Kirmitzoglou, I. & Promponas, V.J. LCR-eXXXplorer: a web platform to search, visualize and share data for low complexity regions in protein sequences. Bioinformatics 31, 2208-2210 (2015).
56. Hoefig, K.P. et al. Eri1 degrades the stem-loop of oligouridylated histone mRNAs to induce replication-dependent decay. Nature structural & molecular biology 20, 73-81 (2013).
57. Grosche, A. et al. The Proteome of Native Adult Muller Glial Cells From Murine Retina. Mol Cell Proteomics 15, 462-480 (2016).
58. Feederle, R. et al. Generation of Pax1/PAX1-Specific Monoclonal Antibodies. Monoclon Antib Immunodiagn Immunother (2016).
59. Sysoev, V.O. et al. Global changes of the RNA-bound proteome during the maternal- to-zygotic transition in Drosophila. Nat Commun 7, 12128 (2016).
Figure legends
Figure 1: Icos responds to inputs from several post-transcriptional regulators. a, d, g,
j, Bar diagrams of Icos mean fluorescence intensities (MFI) over a five-day period of T cell
32 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
activation illustrating changes induced by the CD4+-specific, induced depletion of the
respective genes. Significance was calculated using the unpaired t-test (two-tailed) with n=3.
b, e, h, k, Representative histograms of Icos expression at the specified days. c, f, i, l,
Histograms demonstrating successful depletion of the respective target proteins. o-r,
Western blots showing patterns of dynamic RBP regulation after CD3 and CD28 CD4+ T cell
activation. ns = not significant; * significant (p value = 0.01 – 0.05); ** very significant (p
value = 0.001 – 0.01); *** very significant (p value = < 0.001).
Figure 2: The CD4+ T cell RBPome of polyadenylated RNAs. a, Schematic illustration of
the RNA-interactome capture (RNA-IC) method that was carried out to identify RBPs from
mouse and human CD4+ T cells. b, c, Volcano plot showing the -log10 p-value plotted
against the log2 fold-change comparing the RNA-capture from crosslinked (CL) mouse CD4
T cells (b) or human CD4+ T cells (c) versus the non-crosslinked (nCL) control. Red dots
represent proteins significant at a 5% FDR cut-off level in both mouse and human RNA-
capture experiments and blue dot proteins were significant only in mouse or human,
respectively. d, Enrichment analysis of GO Molecular Function terms of significant proteins
in mouse or human RNA capture data. The 10 most enriched terms in mouse (dark blue)
and the respective terms in human (light blue) are shown. The y-axis represents the number
of proteins matching the respective GO term. Numbers above each term depict the adjusted
p-value after Benjamini-Hochberg multiple testing correction. e, Distribution of IDRs in all
Uniprot reviewed protein sequences (black line), in proteins of the mouse EuRBPDB
database (green line) and in proteins significant in the mouse RNA-IC experiment (red line).
The same plot is shown for human data at the bottom. According to Kolmogorov-Smirnov
testing the IDR distribution differences between RNA-IC (red lines) and all proteins (black
lines) are highly significant in mouse and man and reach the smallest possible p-value
(p<2.2x10-16).
Figure 3: The global RNA-bound proteome of CD4+ T cells. a, Schematic overview of the
OOPS method 39 with phase partitioning cycles increased to five. b-c, Volcano plots showing
33 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
the -log10 p-value plotted against the log2 fold-change comparing the organic phase after
RNase treatment of the interphase of OOPS experiments of crosslinked mouse CD4+ T cells
(b) or human CD4+ T cells (c) versus the same non-crosslinked sample. Red dots represent
proteins significant at a 5% FDR cutoff level in both mouse and human OOPS experiments
and blue dots proteins significant only in mouse or human, respectively. d, Enrichment
analysis of GO Molecular Function terms of significant proteins in mouse or human OOPS
data. Enriched terms are depicted exactly as described for RNA-capture data in Figure 2d. e,
Distribution of intrinsically disordered regions in all Uniprot reviewed protein sequences
(black line), in proteins in the mouse EuRBPDB database (green) and in proteins significant
in the mouse OOPS data (red line). The same plot is shown for human data at the bottom.
According to Kolmogorov-Smirnov testing the IDR distribution differences between RNA-IC
(red lines) and all proteins (black lines) are highly significant in mouse and man and reach
the smallest possible p-value (p<2.2x10-16).
Figure 4: Defining the mouse and human CD4+ T cell RBPomes.
a, Relative abundance (Log2) of proteins identified in a single-shot mouse proteome (orange
dots) plotted by their rank from highest to lowest abundant protein. RNA-binding proteins
detected by RNA capture (top plots) or by OOPS (bottom plots) are highlighted as blue
diamonds. b, Same plots as shown in (a) for human data. c, e, The recently established
database EuRBPDB was used as a reference for eukaryotic RBPs to determine the numbers
of canonical and non-canonical RBPs identified by RNA-IC or OOPS on mouse (c) or human
(e) CD4+ T cells. d, f, The occurrence of RBDs in RNA-IC and OOPS-identified RBPs was
analyzed in comparison with the ten most abundant motifs described for the mouse (d) or
human (f) proteome. g, Venn diagram using four datasets for RBPs in CD4+ T cells as
determined by RNA-IC and OOPS in mouse and human cells. *The core RBPomes contain
proteins present in at least two datasets.
Figure 5: Stat1 and Stat4 are RBPs with regulatory potential. a, b, Semi-quantitative
identification method for RBPs as in 59. In short, GFP-fused proteins were transfected into
34 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
HEK293T cells, crosslinked or left untreated and subsequently immunoprecipitated using an
anti-GFP antibody. The obtained samples were divided for protein and RNA detection by
Western or Northern blotting, respectively. c, Schematic representation of the tethering
assay that was used to investigate a possible influence of the genes of interest (GOI) on
renilla luciferase expression. The affinity of the λN peptide targets the respective fusion
protein to boxB stem-loop structures (5x) in the 3’-UTR of a Renilla luciferase gene where it
can exert its function, if exciting. d, FACS plots using the new monoclonal antibody 4F10
demonstrate specific λN detection of a λN-GFP fusion protein expressed in 293T cells. e,
Western blot showing the expression of the indicated λN-GOI proteins in 293T cells after
transfection. f, Tethering assay results performed in HeLa cells as explained in (c). Two
negative controls were implemented, using constructs without boxB sites or λN expression
without fusion to a GOI. Each measurement was performed in triplicate and was
independently repeated at least twice (n=2).
Figure 6: Identification of proteins in proximity to Roquin-1 in CD4+ T cells. a,
Schematic overview of the BioID method showing how addition of biotin to the medium leads
to the activation of biotin, diffusion of biotinoyl-5’AMP and the biotinylation of the bait
(Roquin-1) and all preys in the circumference. b, Equimolar amounts of protein were loaded
onto a PAGE gel for Western blotting applying an anti-biotin antibody. Efficient biotinylation
of both baits BirA*-Roquin-1 and BirA*-GFP (control) could be demonstrated. c, Histogram
showing that transduction of CD4+ T cells with retrovirus to inducibly express BirA*-Roquin-1
lead to the efficient downregulation of endogenous Icos. d, Identified preys from Roquin-1
BioIDs (n=5) in CD4+ T cells. Depicted are all significantly enriched proteins with the
exception of highly abundant ribosomal and histone proteins. Dot sizes equal p-values and
positioning towards the center implies increased x-fold enrichment over BirA*-GFP BioID
results. e, Venn diagram showing the overlap of RBPs from the CD4+ T cell RBPome with
the proteins identified by Roquin-1 BioID in T cells. f, Listed are all 38 proteins (59%) that
35 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
are Roquin-1 preys and RBPs. Colored fields indicate proteins that were chosen for further
analysis.
Figure 7: Higher order Icos regulation by Roquin-1 and Celf1. Treatment with 4’-OH-
tamoxifen of CD4+ T cell with the genotypes Rc3h1fl/fl;Rc3h2fl/fl;rtTA3 without (WT) or with the
CD4Cre-ERT2 allele (iDKO) were used for transduction with retroviruses. Expression levels
of Icos and four additional Roquin-1 targets in WT and iDKO cells were analyzed 16h after
individual, doxycycline-induced overexpression of 46 GFP-GOI fusion genes. a, c, e, g, The
geometrical mean of each GOI-GFP divided by GFP for each Roquin-1 target was calculated
and the summarized results are shown as bar diagrams for (a) Vav1, (b) Rbms1, (c) Cpeb4
and (d) Celf1. b, d, f, h, Representative, original FACS data are depicted as histograms or
contour plots. Experiments for the negative Vav1 result were repeated twice, those for
Rbms1, Cpeb4 and Celf1 at least three times.
Supplementary Fig. 1: RNA-IC supporting results.
a, Western blots demonstrating specificity of the newly established pan-Ythdf monoclonal
antibody 17F2 for N-terminal GFP fusions of Ythdf1, Ythdf2 and Ythdf3, which are absent
from non-transfected 293T cells. b, Silver staining analysis of oligo(dT) captured samples
with and without UV irradiation. c, Quantitative RT-PCR to determine RNA pull-down
efficiency with (CL) and without crosslink (nCL). Error bars show the standard deviation
around the means of three independent experiments. d, Western blotting of UV irradiated
and nonirradiated samples of EL-4 T cells. Membranes were probed with antibodies for the
known mRNA-binding protein polypyrimidine tract binding protein 1 (Ptbp1) and β-tubulin. e,
Distribution of low complexity regions in all Uniprot reviewed protein sequences (black line),
in proteins in the EuRBPDB database (green) and in proteins significant in the RNA-IC data
(red line). The left plot shows mouse data and the right plot human data. According to
Kolmogorov-Smirnov testing the LCR distribution differences between RNA-IC (red lines)
and all proteins (black lines) are highly significant in mouse (p<2.2x10-16) and man
(p=7.8x10-16).
36 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplementary Fig. 2: RNA-IC on mouse and human iTreg cells. a, Volcano plot
showing the -log10 p-value in relation to the log2 fold-change comparing the RNA-capture
from crosslinked mouse regulatory T cells (left plot) or human regulatory T cells (right plot)
versus the non-crosslinked control. Red dots represent proteins significant at a 5% FDR
cutoff level in both mouse and human RNA-capture experiments and blue dots proteins
significant only in mouse or human, respectively. b, Venn diagram using four datasets to
compare RNA-IC derived RBPomes of effector T and induced Treg cells from mouse and
man.
Supplementary Fig. 3: OOPS supporting results. a, Agarose gel demonstrating
disappearance of the typical 18S/28S rRNA bands after crosslinking and appearance of
upshifted protein-RNA adducts (black arrows) in MEF cells with and without doxycycline-
induce Roquin-1 expression. rRNA bands reappear after proteinase K (PK) treatment (grey
arrows) and after RNase treatment the protein-RNA adducts in the wells disappear. b,
Western blots showing that known RBPs, such as Roquin-1 and Gapdh can be detected in
interphases, in the case of Roquin-1 only after induced expression. * cleavage product. c, d,
Volcano plot showing the -log10 p-value plotted against the log2 fold-change comparing the
interphase of OOPS experiments of crosslinked mouse CD4+ T cells versus the organic
phase after RNase treatment of the interphase. Glycoproteins are highlighted in red.
Enrichment analysis of GO Molecular Function terms was performed for proteins with a log2
fold-change larger than 2. The 20 most enriched terms are depicted.
Supplementary Figure 4: Gene ontology enrichment analysis. Enrichment analysis of
GO Biological Process and GO Cellular Component terms of significant proteins in mouse or
human RNA-IC data (top row) or OOPS data (bottom row). The ten most enriched terms in
mouse (dark blue) and the respective terms in human (light blue) are shown. The y-axis
represents the number of proteins matching the respective GO term. Numbers above each
term depict the adjusted p-value after multiple testing correction (Benjamini-Hochberg).
37 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Supplementary Fig. 5: All currently annotated CCCH and KH domain containing RBPs
and their detection in the RNA-IC and OOPS-identified RBPomes. All RBPs with the
indicated RBDs are listed according to EuRBPDB. Colored boxes indicate significant
enrichment by RNA-IC or OOPS in mouse or human CD4+ T cells.
Supplementary Fig. 6: Identification of Roquin-1 preys by BioID in MEF cells. a,
Western blots showing doxycycline-induced expression of Myc-BirA*-Roquin-1 or Myc-BirA*
in MEF cell clones. b, FACS blot demonstrating that as in T cells (Fig. 6c) N-terminal fusion
of Myc-BirA* to Roquin-1 does not affect its function and c, the fusion protein can biotinylate
Roquin-1 and d, its cofactor Nufip2. e, Identified preys from Roquin-1 BioID in MEF cell
clones. Depicted are 55 of 143 significantly enriched proteins (n=4) for better comparison
with Fig. 6d. Yellow dot color indicates identification of the Roquin-1 prey in MEF and T cells.
Dot sizes equal p-values and positioning towards the center implies increased x-fold
enrichment.
Supplementary Fig. 7: Overlap between Roquin-1 preys in MEF cells and the CD4+ T
cell RBPome. a, Venn diagram showing that 96 proteins (67%) are Roquin-1 preys and also
RBPs in T cells. b, Table listing these 96 proteins. Colors indicate which genes were clone
for downstream experiments and if they were identified by MEF BioID only (red) or
additionally in the T cell BioID.(blue).
Supplementary Fig. 8: Supporting results for higher order Icos regulation by Roquin-1
and Celf1. a, Microscopical images showing different localizations of the GFP signal as a
result of GFP-GOI subcellular targeting in transfected 293T cells. b, Schematic
representation of the experiment performed in Fig. 7. c, Treatment with 4’-OH-tamoxifen of
CD4+ T cell with the genotypes Rc3h1fl/fl;Rc3h2fl/fl;rtTA3 without (WT) or with the CD4Cre-
ERT2 allele (iDKO) were used for transduction with a retrovirus expressing GFP-Roquin-1.
Expression levels of the Roquin-1 targets Icos, Ox40, Ctla4 and IκbNS were analyzed and
work as a positive control for the experiment. d, e, Cells taken from an experiment as
performed (b). qPCR (d) and Western blot (e) performed on cDNA or protein lysates,
38 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
respectively, derived from CD4+ T cells (WT) after doxycycline-induced expression of the
indicated fusion proteins.
39 a b c CD4-Cre-ERT2 (WT) Rc3h1-2f/f;CD4-Cre-ERT2 0d 4d 5 ** ns ** *** ** 100 100
) bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint
10 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 4 60 60
3 20 20 Icos (MFI log Cells (%) Cells (%) 2 3 3 4 5 3 3 4 5 3 3 4 5 0d 1d 2d 3d 4d 5d -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 Icos-FITC Roquin-AF647 d CD4-Cre-ERT2 (WT) e f Zc3h12f/f;CD4-Cre-ERT2 0d 4d 5 ns ns ns ns ns ns 100 100 ) 10 4 60 60
3 20 20 Icos (MFI log Cells (%) 2 Cells (%) -103 0 103 104 105 -103 0 103 104 105 -103 0 103 104 105 0d 1d 2d 3d 4d 5d Icos-FITC Regnase-1-AF647 g CD4-Cre-ERT2 (WT) h i Wtapf/f;CD4-Cre-ERT2 1d 5d 5 100 100 ns ns ns ns ns
) ** 10 4 60 60
3 20 20 Icos (MFI log Cells (%) 2 Cells (%) 3 3 4 5 3 3 4 5 3 3 4 5 0d 1d 2d 3d 4d 5d -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 Icos-FITC Wtap-FITC j CD4-Cre-ERT2 (WT) k l Dgcr8f/f;CD4-Cre-ERT2 1d 5d 5 100 ns ns ns ns * ** 100 ) 10 4 60 60
3 20 20 Icos (MFI log Cells (%) 2 Cells (%) 3 3 4 5 3 3 4 5 3 3 4 5 0d 1d 2d 3d 4d 5d -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 Icos-FITC Dgcr8-APC
m 0d 1d 2d 3d 4d 5d kDa n 0d 1d 2d 3d 4d 5d kDa 63 Roquin-2 TTP Roquin-1 135 100 Pan-Ythdf 75 75 Gapdh 35 579 510 63 o 0d 1d 2d 3d 4d 5d Regnase-1 75 Rbms1 48 Pan-Ago 100 Gapdh 35 0d 1d 2d 3d 4d 5d Nuf p2 75 p Cpeb4 75 Fmrp 75 Gapdh 35 Fxr1 75 q 0d 1d 2d 3d 4d 5d Celf1 Fxr2 75 48 Gapdh 35 Gapdh 35
Figure 1 a Differentiation UV crosslink Oligo(dT) Mass
into Teff (254 nm) capture Spec. bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowedAAAAA without permission. TTTTT Cell lysis Elution CD4 T cell Expansion + oligo(dT) RNase isolation beads digest
Poly(A)-mRNA
b c mouse RNA-IC human RNA-IC 7 8 SYNCRIP significant in mouse significant in human HNRNPC Pum2 7 6 and human G3bp1 Hnrnpm and mouse HNRNPL Hnrnpab Tial1 EIF3A significant in mouse Raly 6 significant in human HDLBP 5 Pum1 Ptbp3 DDX3X CSDE1 Rbms1 Mbnl1 5 RALY 4 Elavl1 ELAVL1 Celf1 4 CELF2 RC3H1 CELF1 3 Rc3h1 Cpeb4 RBMS1
3 p-value [-log10] p-value [-log10] 2 Vav1 Crip1 2 Ldha CPEB4 1 1
0 0
-10 -5.0 0 5.0 10 -10 -5.0 0 5.0 10 Student's T-Test Difference CL vs nCL [log2] Student's T-Test Difference CL vs nCL [log2]
d e IDR - RNA-IC 1.0 GO - Molecular function mRNA-IC mEuRBPDB 0.8 all proteins
80 2.1E-44 3.9E-40 0.6 70 mouse human 0.4 60
50 0.2
proportion of proteins
3.2E-18
4.4E-20
1.9E-23 1.1E-21 40 0.0
# of proteins
9.1E-11
4.1E-8
5.5E-10
4.1E-8
2.5E-10
3.9E-10
1.2E-7
6.7E-10
5.8E-8 9.1E-11
1.5E-8 3.9E-10 3.1E-9 30 5.0E-9 1.0 hRNA-IC hEuRBPDB 20 0.8 all proteins 10 0.6 0 0.4
0.2
mRNA binding miRNA binding proportion of proteins
poly(U) RNA binding 0.0 mRNA 3'-UTR binding regulatorypoly-purine RNA binding tract binding 0.0 0.2 0.4 0.6 0.8 1.0 poly-pyrimidine tract bindingtranslation regulator activity amount of region single-stranded RNA bindingdouble-stranded RNA binding
Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. a b mouse 36 human 36 32 32 28 RNA-IC bioRxiv preprint28 doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights24 reserved. No reuse allowed without permission. [log2] 24 20 20
Rel. abundance 16 500 1500 2500 3500 500 1500 2500 3500 Rank Rank 36 36 32 32 OOPS 28 28
[log2] 24 24 20 20 Rel. abundance 500 1500 2500 3500 500 1500 2500 3500 Rank Rank = identified in single-shot proteome = significant in RNA-IC or OOPS
c d mEuRBPDB 4096 128 mOOPS 1024 64 mRNA-IC 256 32 64 16
16 8 # of RBPs #of # of RBPs of # 4 4 2
1 1
RIO1
KH_1
ResIII
zf-met DEAD
RRM_1 RRM_5 zf-C2H2
e f zf-CCCH MMR_HSR1 4096 hEuRBPDB 128 hOOPS 1024 64 hRNA-IC 256 32 64 16 8
16 # of RBPs of # # of RBPs of # 4 4 2
1 1
KH_1
ResIII
zf-met DEAD
RRM_1 RRM_5
zf-C2H2
zf-CCCH
MMR_HSR1 Methyltransf_31 RBD abundance in proteome g mOOPS hOOPS 1255 1159
mRNA-IC 519 424 hRNA-IC 312 308 11 486 14
35 23 27 24
8 161 8 16 20
Mouse RBPome: 1352 38 Human RBPome: 1250 *Core RBPome: 798 *Core RBPome: 801 Figure 4 a bioRxiv preprintUV doi:- https://doi.org/10.1101/2020.08.20.259234+- + - + ; this- version+ posted August 20, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Roquin
GFP
Rbms1
Ldha
Protein mRNA b UV - + - + - + - + Stat1
Stat4
Crip1 0,2% Input 8% Capture 0,014% Input 92% Capture
Protein mRNA c d GFP λN-GFP 105 λN-GOI 104 ? 3 anti-λN
λ N 10 (4F10) Renilla Firefly poly(A) luciferase luciferase poly(A) 102 Promoter BoxB sites Promoter 101 101 102 103 104 105 GFP e f
no boxB 120 5 boxB
λN-Roquin-1(148λN-Vav-1(111λN-Crip1(22 kDa)λ N-Stat1(104kDa) λkDa)N-Stat4(101λ N-Celf1(65kDa)λ N-Rbms1(58kDa) kDa) λN-Cpeb4 kDa) λN-Pat1b (93 kDa) (108 kDa) kDa 135 80 100 RLU 75 anti-λN 63 (4F10) 40 48
35 0 25
anti-Gapdh
Figure 5 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
(10x) (10x)
a b qi (10 oquin BirA*- (10 oquin
BirA-GFP BirA-GFPBirA-Roquin BirA-RoquinBirA-san BirA-san r-a (10x) irA-san
Dox-inducible BirA* fusion: Labelling radius BirA*-GFP Roquin-1 (10x) irA-san
(10x) (10x)
ID ID ut
(~ 10 nm) ut
oDBirA-GFP ioID oDBirA-GFP ioID
APd APd
APd APd
Input Input
Input Input
iI BirA-R BioID iI BirA-R BioID
Bio Bio
B B
Inp Inp
BioID BioID
BioID BioID
iI B BioID iI B BioID
APd APd
Input BioID FT Input BioID FT BioID BioID addition BirA*- of biotin 135 135 Roquin-1 anti-biotin BirA* Roquin-1 BirA* Roquin-1 100 100
75 75 BirA*-GFPn n
ti ti
o o
Bi Bi Roquin-1-binding proteins c Transduced T cells 100 Peripheral proteins BirA*- Biotinoyl-5‘AMP 80 Roquin-1 BirA*-GFP Covalently bound biotin 60 40 20
d Cells (%) Nufip2 Stat3 0 103 104 105 Ywhab Actb Ubap2 Icos-PE Sash3 Ubap2l Llph Helz Surf6 Pdcl3 Alyref Atxn2l Fam120a R3hdm2 Rc3h1 = bait Coro1b Pcca Pspc1 Cnot2 Prcc2b/a/c Cnot1 p < 0.001 Upf1 Plec Psmc2 Swap70 Rc3h1 p < 0.002 Rc3h2 Keap1 p < 0.004 Zap70 Tdrd3 100x Rrbp1 Ybx1 Ssb p < 0.02 Eif4enif1 Abcf1 Lsm14a 50x Ythdf3/1 p < 0.05 Rtn4 Mccc1 Pip4k2c 20x Tnrc6a/b Fnbp1 10x Pc Dmnt1 Gapdh Xrn1 5x Ebna1bp2 Ddx6 Edc4 Usp36 2x f Smg7 Roquin-1 T cell BioID overlap with RBPome
Expanded T cell Abcf1 Fnbp1 Prrc2c Tnrc6b e RBPome (1320) Alyref Helz Pspc1 Ubap2 Atxn2l Hist1h1c R3hdm2 Ubap2l Cnot1 Hist1h3b Rc3h1 Upf1 T cell Roquin-1 Cnot2 Lsm14a Rpl23a Xrn1 BioID (64) Ddx6 Nufip2 Smg7 Ybx1 1282 3826 Ebna1bp2 Pc Ssb Ythdf1 Edc4 Plec Stat3 Ythdf3 Eif4enif1 Prrc2a Tdrd3 Fam120a Prrc2b Tnrc6a cloned for induced expression
Figure 6 WT iDKO WT iDKO a Vav1 b Icos Icos 2,5bioRxiv preprint doi: https://doi.org/10.1101/2020.08.20.259234; this version posted August 20, 2020. The copyright holder for this preprint (which was notWT certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 2,0 iDKO Icos-PE
1,5
100 5 1,0 Ctla4 Ctla4 10 80 4 10 0,5 NS 60 103 Icos Ctla4 Ox40 I κ B Cells (%) 40 2
Ctla4-PE 10 0,0 Mean Fluorescence (GOI/GFP) 20 1 Regnase-1 10 0 101102103104105 101102103104105 c Rbms1 d Icos Icos 2,5 WT 2,0 iDKO Icos-PE
1,5
1,0 100 Ctla4 Ctla4 105
80 104 NS 0,5 60 3
Icos 10 Ctla4 Ox40 I κ B
Cells (%) 40
Ctla4-PE 2 0,0 10 Mean Fluorescence (GOI/GFP) 20 Regnase-1 101 0 101102103104105 1 2 3 4 5 e Cpeb4 f 10 10 10 10 10 3,0 Ox40 Ox40 WT 2,5 iDKO
2,0 Ox40-APC
1,5 100 5 1,0 Ctla4 Ctla4 10 80 104 0,5
60 103 NS
Cells (%) 40 2 Ctla4-PE Mean Fluorescence (GOI/GFP) 0,0 Icos 10 Ctla4 Ox40 I κ B 20 101 0 5 Regnase-1 1 2 3 4 5 g Celf1 h 10 10 10 10 10 10110210310410 Icos Icos 2,5 WT 2,0 iDKO Icos-PE
1,5
100 5 1,0 Ctla4 Ctla4 10
80 104 0,5 NS 60 3
Icos 10 Ctla4 Ox40 I κ B
Cells (%) 40 2 0,0 Ctla4-PE 10 Mean Fluorescence (GOI/GFP) 20 1 Regnase-1 10 0 101102103104105 101102103104105 target expression GFP
WT+GFP-GOI iDKO+GFP-GOI T cell+GFP
Figure 7